Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Fastest1 on November 28, 2015, 11:07:18 AM
-
I am having an issue quantifying just what is responsible for smooth and fast motion on a PP based PC.
I have always used computers that met the minimum requirements of Mach 3. Some PC's propel my machines to seemingly unuseable speeds and others stall, same controllers, same settings in Mach. Just different PC's.
I have used the optimization steps in Mach and checked and or changed my port status etc.
Many times during the pursuit of speeds attained prior I would hear a clicking coming from the steppers.
I have only read a few other threads that mention this. In my case it always sounded mechanical and had me adjusting gibs and bearings, lubricating, cleaning etc. Never to improve.
Just lately I was finally smart enough the remove the steppers from my lathe only to find the sound was still there. I drove the axis by hand and was amazed at the smoothness and lack of effort required to move any of the axis'.
Friction is not causing my stalling issues. Nor are the speeds I am asking of the steppers.
Changing the acceleration has no effect nor turning the toolpath on or off.
Searching the forums for a specific mobo, RAM, graphics card with guaranteed success almost impossible. Is it the software? The parallel port itself? A graphics card? I cant imagine it is luck of the draw.
I realize that I could use a motion controller of some flavor. I have a few. However my stubbornness makes me pursue WHY? As mentioned above I have had great success with the PP. No need to change if I can find out why.
Here are 2 different screenshots of the DriverTest results, not mine just pics from the net. Looking at the results of the tests and the signals. Which would be the better computer to use and why?
Both are testing at the same speed.
Both claim system excellent.
-
I personally think that Drivertest, only gives you a rough idea of what to expect. It either fails, or says System Excellent.
I think that the main factor in pulse stream quality is the motherboard.
And to really do a good comparison, you need to look at the pulses with an oscilloscope.
-
Is there a particular mobo that you are familiar with that performs better than another?
Sent from my iPhone using Tapatalk
-
No, I've only really used one.
A really old Dell GX150. I have a spare if it ever goes bad.
They're about $20-$30 on Ebay with a 1Ghz PIII. You do need the appropriate Dell power supply to run them, as they use a different pinout than standard.
-
Your driver test shows me that your PP is not that great or there is interference.
The reported pulses per Second should be Very Close to the selected Kernal speed ie; 25K & you have 23.984K. The closer the better.
You should not see that manny and heigth of pulses in the graph. Should or could have a few /short/ sproadic and almost just a straight line.
That said, I think you signal out of the PP is crappy and you hear it via the clicking / ticking sounds. Probably noise on the signal and would need to look at the signal with a scope.
You can't clean up a bad signal when it's generated by the PC to the PP.
All of the Dell's I had were crappy for PP use.
My thoughts......
RICH
-
Rich, I appreciate the reply. These driver test results were images I got off of the net and not actually mine. That being said, mine look no better and have plenty of spikes too.
I did a fresh install of XP today and it did seem to remove the click of the stepper. I gained no speed though.
Rich, did you ever add a parallel port card to yours or were they always the stock Dell PP card?
Sent from my iPhone using Tapatalk
-
The DEll's we tested were all the same and required an external motion device. Tried adding a non stock PP card to the Dell but the scope
still showed ringing on the pulses which created bad stepper pulsing. So I used a SS for a while with it, but they were limited at the time and didn't satisfy my needs. One can try to filter / clean up the pulse signal but not as easy as it sounds ( basically bad PC generated signal and it will carry on through the system). The Dell
pulse rates were bad ie; 25k kernel and they would vary down to as low as say 14.5K.
My solution was to make up a new PC for just CNC and the Driver test reported pulsing is +- 3 as compared to the selected Kernal speed and graph is straight lined.
Use a RoseWill dual PP card in the dedicated pc.
BTW, I added a DRO for the pulse rate to the Diagnostics screen.
RICH
-
One thing that's critical with my Dell is to install windows in Standard mode, rather than ACPI. This is done in the bios , preferably before installing windows.
This setting is the difference between an unusable PC and a near perfect pulse train.
On my Dell, the pulse rate is very close to 25,000, and only varies by about 2 pulses while running. I don't recall the exact numbers, but it's something like 24890-24892.
In ACPI mode, motors will not turn smoothly at all. Constantly stuttering and stopping.
-
Another important factor, if you have a multiple core processor, is to disable either C1E or EIST in the bios.
-
Good tip Gerry. Don't recall what models we tested and it also include laptops.
Don't even recall what mode they were in since it's been a long time since we did the testing.
RICH
-
That gives me plenty to pursue.
Thanks for all of the input.
Sent from my iPhone using Tapatalk
-
This is challenging to say the least. Apparently more than 1 of us do agree they have experienced different results on similar computers.
In my case I have proof of my machine running 330ipm in all 3 axis during a warm up routine. Yet now, on a different computer(s) fresh installs of everything. No chance I could get close. Maybe somewhere around 180ipm at the max and of course divide ipm by 3 for the actual single axis speed.
What is more interesting is trying to find the information on what is responsible for the difference. Nobody seems to have any idea other than to get another computer, that might or might not have the same issue.
Not being very computer knowledgeable but just dumb enough to try anything at this point, registry edits, BIOS setting changes, Reinstalling XP in every way you can, while holding F7 attempting to disable the APIC (not ACPI and there is a difference) or whatever. My theory is if it is broken, how much more broken are you going to make it? None, its broken.
Then you read, disable this in BIOS etc. only to find your mobo doesnt have a APIC disable for example. Hackers and overclockers tell you how to use modbin6.exe to do some form of editing, hallelujah there wasnt an instructional on that. Or I would probably be complaining about a BSOD on another forum I had to become a member of ;-)
I soon start calling anybody and everybody I know who might know how to solve or investigate this issue. At one point I am speaking to a computer tech from a repair center whose father runs Mach 3 (total coincidence). He almost takes it as a personal challenge. After he explains more concepts than I can comprehend he keeps referring to it being a software problem, not hardware. Specifically an update on the Windows system. This update creates some form of polling that causes the interrupts. He apparently finds articles referring to it (not a Mach specific problem). He says the issue was most likely caused by a wdsclient.dll and refers to a number 935772. Again both over my head. He claims the problem was caused by this specific update. Just as you think there is a light at the end of the tunnel, he says "the problem will be finding the correct .dll that did work". As maybe there is no library of options?
Being hesitant about throwing money at a problem, you seem to read a lot about how modern or small computers might not work. Then people here have good experiences with a pico or mini itx too.
I had thought the APIC had some bearing on the PP output and its noise. Then I see a driver test of a modern mini board and it almost looks flat. The green APIC LED is lit? How can this be? So of course now just more confusion. With a flat line or seemingly no or very few interrupts, would this allow me to experience the speeds and performance I once had? The saga continues.
-
Why not just spend $100 on a UC100 and abandon the parallel port?
-
Stubborn?
In reality I have had great experiences in the functionality of Mach from the beginning. Not saying I havent had problems but they were usually my faults as I was GREEN. I have gained a bit of knowledge and experience along the way (and you have contributed to it Gerry). For my needs and intelligence level the PP was exceptional when the PC was working correctly, right M3 version etc.
I have tried both USB and ESS versions of the SS over the years only to find a limitation or 2 that I cant seem to work around. Or dont like. Whether it be lack of BC or threading or the constant faulting out. (Maybe the UC would be the same?) Really just another level of learning, another item to make it more complicated. I was trying the KISS principle. The latest is the homing issue in Turn that inspired this rant. Whether it is an ESS or Mach issue. It doesnt happen for me like than in Mill on a LPT. I dont think my need was ever for more pulses. Maybe on the Dyna's at 101000 steps per unit. Now I am in the 8000 steps per unit so quantity doesnt seem needed.
I guess I really want to use a late model mini with an LPT if it will work. If these problems are in the computer are they just limited to the PP? Or do these glitches and interrupts also effect the performance via a different motion controller only at a higher speed?
-
For all practical purposes, the parallel port is the computer, especially if it's a built in port.
I dont think my need was ever for more pulses.
It's not about more pulses. Just better pulses.
The problem is that the PC, for whatever reason, can't generate smooth pulses.
So eliminate the problem, and get a device that's specifically designed to generate smooth pulses. A motion controller. The PC should have nothing to do with the pulses generated by the motion controller.
-
That is what I did with the ESS. It got me to this point because after all of the configuration. Satisfying performance and all. I cant get the machine to home correctly. No matter what I do to rectify it. It doesnt fix it. Though I have and can run machines without limits or home switches. I dont want to. Once you have experienced machines with repeatable homing, you dont want to go back.
Nobody on the SS forum or the Mach forum knows why or which piece is dominant for the dialog box reverting after hitting the reset buttons for example. Not including why the homing offset value in the homing/limits page doesnt work. Maybe it is Turn? But it doesnt work like this on a PP with Mill. If I could get around the 1 homing issue I would move on. Without it, its a deal killer.
Regarding the interrupts and whether they transfer thru the motion controller was just a thought, not based on anything but suspicion. Just seems like the same stutter would still be there. Plenty to be wrong about.
-
Hmm..........
What you need is my lathe screen set. I don't use switches and frankly don't see any need for them on my lathe.
I use a quick change for the all the tools ( maybe I'll have a turret some day, good grief another project ) and use probing to find the lathe X & Z zero, manually set a home position to where it is usable for the current job, and probe to set the G54 offset in Z. I can probably do that as fast as having switches to set up a job!
Guess it just a matter of how one works.
One of these days the screen manual will be done and then I'll post it, but, until I take time away from some other interests the screen set will just need to remain in the background.
Toot Toot, Bang Bang, Chop Chop, etc, ::)
which means: Music , Guns , Knife Making, ( and other stuff which can't be said) :)
Got to go and prepare for Santa's arrival as the bigger chimney is not finished, ;D
RICH
-
My lathe has turret.
If you had home switches you would like them. I have done both ways. I like soft limits too. Especially when they are based off of a homing switch instead of a random point.
-
Rich that is not a screen it is a NOVEL ;D
You do not need switches with steppers as you can simply bump them up against a hardstop then set home. Now don't try that with servos (;-)
(;-) TP
-
Now don't try that with servos (;-)
(;-) TP
That is the way I home my turret and it has a servo ;)
Hood
-
Not a good idea with a servo it is a good way to blow out the final drive section of the drive. To touch off is one thing but to drive it hard into the stop is another. I am afraid that most don't know the difference.
(;-) TP
-
Don't think it will do my drive much harm TP ;)
It specifically allows you to do so, see pic., just set a low current value whilst homing :)
Hood
-
Ok I am back.
Over the last few weeks I obtained an Advantech PPC 174T, combo 17" touchscreen and 2.66ghz processor with 2g ram (brand new with the plastic film still on). It came with the hard drive wiped clean, no OS. Ebay is a wonderful thing! Less than 300 to my door.
I loaded a copy of Xp Pro SP3, Mach 3 and that is it. Well I did DL and install the drivers from Pentouch and Advantech.
I connected thru the PP to my MX3660. Tuned it to 45khz and cant stall my steppers at 200 ipm! They do stall at the 280 ipm range with an accel value above 70. This is at 8250 steps per unit (inch). This is on the machine in operation. I was stalling before at 60-70 ipm on the machine or not.
I am not using a motion control board such as the ESS or Pokeys. In fact I am getting speeds higher than the ESS with the old computer. No ticking of the steppers etc. Quite odd.
I could not disable the APCI when installing XP no matter what I tried. F5 and or F7 had no effect on the installation. Actually F5 gave me an option of standard install but selecting it created an endless boot loop. Whether I did the change in BIOS or thru device manager.
I didnt do any of the optimization steps except disabling updates and eliminating the remote assistance. All seems to be working great. I will not be going online with this PC. Also the driver test is a little better looking (less spikes) but nowhere near a smooth line. Baffled but happy.